Complete Amino Acid Sequence and in vitro Expression of Rat NF-M, The Middle Molecular Weight Neurofilament Protein

Transcription

1 The Journal of Neuroscience, August 1987, 7(8): Complete Amino Acid Sequence and in vitro Expression of Rat NF-M, The Middle Molecular Weight Neurofilament Protein Eugene W. Napolitano, Steven S. M. Chin, David Ft. Colman, and Ronald K. H. Liem Departments of Pharmacology and Cell Biology, New York University School of Medicine, New York, New York A Xgtll expression library was prepared from rat brain and screened with a polyclonal antiserum, which recognizes both NF-H and NF-M. An NF-M cdna clone (pnf-m3c = 1.6 kb) was isolated and characterized. The fusion protein of NF- M3C, when used as an affinity matrix for the anti-neurofilament serum, isolated a subpopulation of antibodies specific for NF-M. Northern analysis demonstrates a single band of approximately 3000 nt and a constant message level for NF-M during postnatal develpoment from postnatal day 0 (PO) to adulthood. Using pnf-m3c as a probe, a second cdna clone was isolated from a Xgtll rat brain expression library (pnf-m2d = 2.7 kb). The 2 clones were sequenced and pnf-m2d was found to encode the entire rat NF-M protein. The calculated molecular weight is 95,600, which is only 65% of the molecular weight determined by SDS-PAGE. The amino acid sequence of rat NF-M shows the conserved rod segment present in all intermediate filament proteins. The molecule also contains an unusual C-terminal extension with stretches of glutamic acid, which could contribute to the anomalous migration of this protein on SDS-PAGE and the fact that NF-M does not readily assemble into filaments. The pnf-mpd clone was transcribed and translated in vitro utilizing a rabbit reticulocyte lysate system. The resulting radiolabeled translation products were unexpectedly shown to comigrate with purified rat NF-M on l- and 2-dimensional gels, even though the translated protein is not phosphorylated. The neuronal cytoskeleton, like the cytoskeleton of most eukaryotic cells, is composed of 3 filamentous networks. Actinbased microfilaments and tubulin-based microtubules constitute 2 phylogenetically highly conserved cytoskeletal systems, whereas the neurofilaments are 1 of 5 independent but related subclasses of intermediate filaments (IFS) exhibiting tissue-specific expression. The preferential expression of the other specific IFS in a particular differentiated cell type occurs as follows: Received Nov. 26, 1986; accepted Feb. 12, This work was supported by grants NS15182, EY03849 and NS E.W.N. was an NIH predoctoral trainee , S.S.M.C. is an MSTP trainee present, and R.K.H.L. and D.R.C. are both recipients of Career Awards from the Irma T. Hirsch1 Foundation. We thank Drs. Lise Bemier, James Salzer, Nick Cowan, Sally Lewis, and Maria Norgard for helpful suggestions and discussion of the work. We also thank Susan Babunovic for her assistance with the photographic work. Correspondence should be addressed to Dr. Ronald Liem, Department of Pharmacology, New York University School of Medicine, 550 First Avenue, New York, NY Copyright Society for Neuroscience /87/ $02.00/O vimentin in cells of embryonic and mesenchymal origin; desmin in all 3 types of myogenic cells-smooth, skeletal, and cardiac; cytokeratins in epithelia; glial filament protein (GFAP) in fibrous astrocytes and Bergmann glia (for review, see Lazarides, 1982). Extensive physicochemical, biochemical, and amino acid sequence analysis have revealed that all IF subunits harbor a conserved 40,000-molecular-weight central hydrophobic oc-helical domain of 310 amino acid residues, which gives rise to the extended coiled-coil polymers characteristic of IFS (Geisler and Weber, 1982). In contrast, the non-a-helical amino- and carboxy-terminal regions of IFS contain extensions that exhibit a high degree of variability in sequence and length (Geisler and Weber, 1983; Geisler et al., 1983). Neurofilaments, by virtue of their neural-specific expression, have served as useful markers for the identification of neuralderived cells and tissues, as well as providing a model system for studying the differentiation of nerve cells during ontogeny (Shaw and Weber, 1982; Willard and Simon, 1983; Pachter and Liem, 1984). In mammals, NFs are composed of 3 distinct polypeptide subunits with apparent molecular weights of 68, ,000, and 200,000 (Hoffman and Lasek, 1975; Liem et al., 1978); however, because oftheir anomalous electrophoretic mobilities on SDS-PAGE (Kaufmann et al., 1984) these proteins are referred to as NF-L, NF-M, and NF-H, to designate low-, middle-, and high-molecular-weight subunits, respectively (Willard and Simon, 1983; Shaw et al., 1984). Within single neurons, NF and their subunit proteins present a nonuniform distribution in that they are concentrated principally in axons and appear in comparatively low abundance in dendrites and perikarya (Hirokawa et al., 1984; Shaw et al., 1984). There is substantial evidence implicating NF as determinants of axon caliber in large rapidly conducting myelinated nerves and structural stabilizers of the axonal cytoskeleton involved in the maintenance of the highly asymmetrical cell forms observed in the nervous system (Morris and Lasek, 1982; Hoffman et al., 1984). Of the 3 subunits comprising NF, only NF-L, the low-molecular-weight subunit, has been convincingly shown to be com- petent in self-assembly to promote the in vitro formation of 10 nm filaments (Geisler and Weber, 1981; Liem and Hutchison, 1982) although studies by Gardner et al. (1984) and our laboratory (Tokutake et al., 1984) indicate that NF-M and NF-H form polymerized filaments in vitro under highly specified experimental conditions. The primary differences between the neurofilament polypeptides are the length and composition of their long non-a-helical carboxy-terminal extensions, which in the 2 high-molecularweight subunits, NF-M and NF-H, contain an abundance of charged residues, i.e., lysine and glutamic acid (Geisler et al., 1983, 1984) along with numerous serine phosphates (Jones and

2 The Journal of Neuroscience, August 1987, 7(8) 2591 Williams, 1982; Julien and Mushynski, 1982; Wong et al., 1984; Carden et al., 1985). Although NF-M and NF-H were considered to interact with the NF-L-based core filament in an associative manner, i.e., peripherally bound to the filament, it appears that these molecules assume a hybrid character in which the hydrophobic amino-terminal sequences participate in filament formation, while the carboxy-terminal tailpieces project from the filament. Studies at the immunoelectron microscopic level on filaments reconstituted in vitro and on NF in situ have shown uninterrupted decoration of the filament by antisera to NF-L and NF-M (Willard and Simon, 1981, Sharp et al., 1982), whereas monospecific NF-H antibodies produced a patchy, discontinuous staining pattern with the majority of the immunolabel concentrated between filaments (Debus et al., 1982; Hirokawa et al., 1984; Liem et al., 1985). This latter result has led to the proposal that NF-H, and specifically the 160,000-molecularweight carboxy-terminal domain, may act as an interfilamentous cross-linker, which joins neighboring filaments and rigidifies the axonal cytoskeleton. The assignment of specific functional roles for IF proteins may be facilitated by the isolation of cdna clones, which could be employed in the study of NF expression in vitro and in vivo. A number of such clones have recently been isolated for other IF proteins, including vimentin (Quax-Jeuken et al., 1983; Zehner and Paterson, 1983), GFAP (Lewis et al., 1984), desmin (Bloemendal et al., 1985; Ngai et al., 1985), epidermal keratins (Hanukoglu and Fuchs, 1982, 1983; Steinert et al., 1983), and NF-L (Lewis and Cowan, 1985, 1986). Here we report on the isolation and characterization of cdna clones encoding the 150,000-molecular-weight subunit, NF-M, of rat neurofilaments. Materials and Methods Construction and screening of a rat brain cdna library. Total RNA from 21-d-old rat brain was prepared by the guanidine hydrochloride method of Chirgwin et al. (1979) and poly(a+) RNA was purified by oligo-dt chromatography (Aviv and Leder, 1972). Some 15 rg of this mrna was used as a template for cdna synthesis. For first strand synthesis, Maloney murine leukemia virus reverse transcriptase (Bethesda Research Laboratories, Bethesda, MD) was used in a reaction cocktail containing 50 mm Tris-HCl (ph 7.5) 75 mm KCl, 10 mm DTT, 3mM MgCl,, 0.75 mm of each dntp s, 10 pg/ml oligo(dt),,_,,, 1 fig RNA. 100 &ml BSA. and 200 units of enzvme (60 min. 37 C). After extraction df-the RNA;/DNA hybrids with slightly alkaline (ph 8) phenol : chloroform (1: 1) and chloroform, and precipitation with 2 M ammonium acetate and ethanol (2 vol), double-stranded DNA was synthesized exactly as described by Gubler and Hoffman (1983) and extracted with organic solvents and precipitated as above. ds cdna (2 pg) was treated (20 min, 37 C) with 2 units of mung bean nuclease (PL-Pharmacia, Uppsala, Sweden) in a buffer containing 50 mm NaCl, 30 mm sodium acetate (ph 5.5), 1 mm ZnCl,, and 3% glycerol. Following phenovchloroform (1: 1) extraction and precipitation with 2 M ammonium acetate and ethanol (2 vol), ds cdna was methylated with Eco RI methylase (New England Biolabs, Beverly, MA), and Eco RI linkers were added in an overnight ligation with T4 DNA ligase. Redundant sequences were digested with Eco RI and ds cdna was then sizefractionated on a Sepharose CL-4B column (PL-Pharmacia, Uppsala, Sweden), and cdnas estimated to be larger than 1.3 kb were ligated into Eco RI-cleaved, alkaline phosphatase-treated bacteriophage Xgtll DNA (Vector Cloning Systems, Tansy, CA) (Young and Davis, 1983). Recombinant DNA was packaged into bacteriophage with a commercially available packaging extract (Vector Cloning System). Upon infection with the recombinant phage, cells from the E. coli Y 1090 (Young and Davis, 1983) were plated out and induced with isopropyl-p-nthiogalactopyranoside (IPTG) to synthesize lacz-fusion proteins. An estimated 3 x 10s recombinant plaques were screened with an anti-nf antiserum that reacts strongly with the rat NF-M and NF-H subunits. Immunopositive clones were rescreened, plaque purified, and amplified to obtain sufficient quantities of phage DNA. cdna inserts were excised from bacteriophage DNA and subcloned into the plasmid vector pgem-2 (Promega Biotech Inc., Madison, WI), as well as Ml3 for further amplification and characterization. Using pnf-m3c as a probe, we screened another 3 x 1 OS recombinant plaques for longer cdnas encoding NF-M. The plaques were replicated onto nitrocellulose and hybridized with pnf-m3c labeled with 3zP by nick-translation (Rigby et al., 1977). After hybridization, the filters were washed to a final stringency of 0.2 x SSC (SSC is 150 mm NaCl, 15 mm sodium citrate) at 68 C and exposed to X-ray film. Positive plaques were diluted into 1 ml of phage dilution buffer (50 mm Tris, ph 7.5, 0.1 M NaCl, 0.2% MgSO,, and 0.01% gelatin), and a portion of the bacteriophage was amplified in a 1 ml culture without purification as described by Lewis and Cowan (1986). The resulting phage DNA was digested with Eco RI and the digestion Droducts were seuarated on a 1% agarose gel and transferred 50 nitrocellulose. The blbts were hybridized with the pnf-m3c insert labeled with 32P by nick translation, washed to a final stringency of 0.2 x SSC, 68 C and exposed to X-ray film. The bacteriophage preparation containing the largest insert, which hybridized with pnf-m3c, was then further purified and subcloned into the plasmid vector pgem-2 and M 13 for further amplification and characterization. Epitope selection. Purified recombinant bacteriophage propagated in E. coli Y 1090 were plated onto LB agar dishes and induced to synthesize /3-galactosidase fusion protein by overlaying nitrocellulose filters presoaked with IPTG (Weinberger et al., 1985). After a 2.5 hr induction period, the filters were removed from the plates, washed with PBS, ph 7.4, at room temperature for 15 min, and then incubated with primary antiserum (diluted in 5% BSA in PBS) overnight at 4 C. Unbound antibodies were aspirated offthe filters, which were subsequently washed 3 x 10 min with PBS at room temperature. Specifically bound antibodies were eluted from the filters with 1 ml of 0.1 M alvcine-hcl. DH 2.5, and neutralized to ph 7.0 with 100 ~1 of 1.5 M T&-buffer, ph 8.8. This affinity-purified immune serum was then tested in a Western blot assay performed according to a previously described procedure (Towbin et al., 1979). Sequencing. Double-digested restriction fragments of pnf-m3c and pnf-m2d were subcloned into bacteriophage vectors M13mp18 and M 13mp 19 according to the forced cloning strategy (Messing and Vieira. 1982). pnf-m3c was restriction-digested with Eco RI-PstI, and pnf- M2D was digested with Eco RI-Pst I and with Eco RI-Xho I. The resulting fragments were ligated with T4 DNA polymerase (New England Biolabs) into Ml 3mpl8 and M13mp19 RFs (New England Biolabs), double-digested with either Eco RI-Pst I or Eco RI-Sal I. The vectors containing the inserts were then used to transform E. coli JM109 and prepared into single-stranded DNA sequencing templates. Sequencing was performed on both strands according to the dideoxy chain termination protocol as described by Sanger et al. (1980). Deoxyadenosine 5 -(Lu-35S-thio) triphosphate (New England Nuclear, Boston, MA) was used as the label incorporated in the dideoxynucleotide sequence reactions (Biggin et al., 1983). The rapid deletion system described by Dale et al. (1985) was used to generate the required subclones for sequencing. T4 DNA polymerase and subcloning primers RD22-mer and RD29-mer were purchased from International Biotechnologies, Inc. Northern blot transfer. Total RNA from rat brain at various different ages prepared as above was electrophoresed through denaturing formaldehyde/agarose gels (Boedtker, 197 1) and blotted onto GENE-SCREEN membranes (New England Nuclear). The RNA blots were hvbridized with agarose gel-purified cdna insert from pnf-m3c labeled with 32P by nick-translation (Rigby et al., 1977). The stringency conditions used for washing the blots are specified in the figure legend. In vitro transcription/translation. pnf-m2d was subcloned in the expression vector pgem-2, which contains both the bacteriophage SP6 RNA polymerase and T7 polymerase promoters in opposite orientations. The transcription was performed in the presence of either the SP6 or T7 polymerase for 1 hr at 40 C following the protocol recommended for the Ribroprobe Gemini system (Melton et al., 1984). The final reaction mixture (11 ~1) contained 2 ~1 of 5 x transcription buffer (200 mm Tris-HCl, ph 7.5,30 mm MgCl,, 10 mm spermidine, 50 mm NaCl), 1 ~1 of 100 mm DTT, 1 ~1 (20 units) RNasin (Amersham, Arlington Heights. IL), 2 ul (10 units) SP6 or T7 nolvmerase (Promeaa Biotec). 1 ~1 of a 10 I&M ribonucleotide triphosphate mixture, and-2 ~1 (2 Lg) plasmid DNA. The mixture also contained 2 ~1 of 5 mm of the nucleotide m7g(5 )ppp(5 )G to cap the synthesized mrna. The reaction mixture was incubated for 1 hr at 40 C. The resulting in vitro transcribed mrna

3 2592 Napolitano et al. * cdna Cloning of Middle Neurofilament Protein D A A A A 3 3c 3 i, ~ cioo 2560 N+W u--c la lb 2 Figure 2. Schematic depiction of the cdna clones compared with the NF-M protein. Top, pnf-m3c, center, pnf-m2d; and bottom, NF-M protein. The numbers denote base pairs and indicate the size of the cdna clones. Coils la, lb, and 2 in the helical rod region of the NF-M molecule are shown in the schematic diagram for the protein. Restriction fragments were obtained with either Pst I (closed triangles) or Xho I (open triangles) and subcloned into Ml3 for sequencing. pnf-m3c spans the region of the NF-M molecule starting in coil la into part of the C-terminal extension, whereas pnf-m2d encompasses the entire coding region of the NF-M molecule. a b C Figure 1. Immunoblot analysis of epitope selected neurofilament antibodies. Fusion protein produced by bacteriophage harboring the NF-M cdna insert was used as an affinity matrix for the isolation of a subpopulation of antibodies preferentially reactive against rat NF-M from an immune serum that initiallv recoanized both NF-M and NF-H with equal intensity on a Western blot. P&ally purified rat NF triplet proteins were resolved on 7.5% SDS-polyacrylamide gels, electrphoretically transferred to nitrocellulose, and reacted with an immune serum. Lane a, Coomassie blue stain of unreacted NF triplet proteins; lane b, immunoblot of unadsorbed anti-nf antiserum; lane c, immunoblot of the antiserum used in lane b after the adsorption to and elution from filters containing NF-M specific fusion protein. Note the very strong NF-M reactivity versus the relatively weak reaction of the affinitypurified antiserum against NF-H in lane c when compared with lane b. H, NF-H; M, NF-M; L, NF-L. Open arrowhead indicates a common degradation product of NF-M. was translated in a rabbit reticulocyte lysate system (New England Nuclear) containing?l-methionine according to the procedure specified by the manufacturer. S-labeled translation products were separated on SDS-PAGE and processed for fluorography (Laskey and Mills, 1975). Immunoprecipitation of the translated polypeptides was performed by the method of Goldman and Blobel(1978) using the anti-neurofilament antibody. To determine the isoelectric point of the translated protein, the translation product was separated by 2-dimensional SDS-PAGE as described elsewhere (Pachter and Liem, 1984). This translation product was mixed with a preparation of rat optic nerve IFS in order to compare the RI of the protein with native NF-M. In addition. the translation was done in the presence of 1 PCi *P-y-ATP, followed by immunoprecipitation to determine whether kinases in the reticulocyte lysate system were capable of phosphorylating the translated protein. Results Isolation of cdnas encoding the NF-M subunit of rat neurojilaments The approach adopted to isolate putative cdna clones encoding the middle-molecular-weight subunit of rat NF polypeptide, NF-M, involved the immunoscreening of a bacteriophage Xgtll expression library (Young and Davis, 1983). To optimize the conditions for selecting an NF clone, a cdna library was constructed from rat brain poly(a+) RNA, a source rich in NF mrna transcripts, and screened with a polyclonal antiserum specific for NF-M and NF-H (Liem et al., 1978). Of 3 x lo5 recombinants analyzed, 3 immunopositive plaques were detected in the primary screening. These were then processed through 3 successive rounds of purification, after which one, NF-M3C, consistently remained immunoreactive. The DNA sequence data obtained indicates that the cdna for NF-M is located entirely within the coding region of its corresponding mrna transcript (see below). This NF-positive cdna insert, now designated pnf-m3c and determined to be 1.6 kb in length, was excised from an amplified phage preparation by restriction endonuclease digestion and subcloned into the plasmid vector pgem-2 (Promega Biotec) and viral vector Ml 3 for further characterization. Using the pnf-m3c insert, we screened an additional 3 x 1 OS recombinants to obtain a longer cdna. With this method, we were able to obtain a second cdna clone, which hybridized with the pnf-m3c insert at high stringency (0.2 x SSC, 68 C) and was found to be 2.7 kb in length. The DNA sequence data obtained for this clone indicate that this cdna spans the entire coding region for NF-M, and the cdna insert (designated pnf- M2D) was isolated from an amplified phage preparation and subcloned into the plasmid vector pgem-2 and the viral vector M13. Epitope selection experiment Confirmation of pnf-m3c as NF-M specific was established by epitope selection (Weinberger et al., 1985). Purified recombinant bacteriophage, following induction with IPTG, generated plaques containing immunoreactive fusion protein, which after transfer to nitrocellulose filters, serve as an affinity matrix for the immune serum of interest. We used the bacteriophage NF- M3C clone to select a specific activity from the antiserum that recognizes both the NF-M and NF-H proteins. The fusion protein of the NF-M3C clone, when used in this capacity, isolated a subpopulation of antibodies from the anti-nf serum with enhanced reactivity for NF-M as compared to NF-H in a Western immunoblot assay (Fig. 1, lane c). Figure 1, lane b, shows that prior to the affinity purification, the antiserum binds avidly to both NF-M and NF-H subunits with no apparent preferential affinity for one NF subunit over the other. This result indicates that the bacteriophage NF-M3C fusion product was capable of

4 The Journal of Neuroscience, August 1987, 7(E) 2593 I IO 20 HSYTLDSLGNPSAYRRVPTETRSSFS GGCClCCAAG AlG AGC TAC ACG CTG GAC TCG CTG GGC MC CCG TCC CCC TAC CGG CGC Cl1 CU\ ACC GAG ACC CGG TCC AGC TTC AGT RVSGSPSSGFRSQ SHSRGSPSTV LSYKRS CGT GIL AGC GGT ICC CCC TCC AGC GGC TTC CGC TCG CAL TCC TGG TCC CGC GGC TCG CCC AGC ACC GTG TCC ICC TCC TAC MG XC AGC 6 71) A A L A ;- R L A Y S S A M L ius A E S S D F S ij- S S S L L N CCC CTC XC CCG CGC CTC GCC TAC AGC TCG GCT ATG CTC AGC TCG GCC CAG AGC AGC C:C CAC TTC AGC CAG TCC TCT TCG CTG CTT MC ID0 110 E E 0 L 0 G R F G Y I E G G S YGDYKLSRS GGC GGC TCC GGC XC CAC TAC MC CTG TCC CGC TCA AiC GAG AiA C4G CAG CTG CAG C&G C:G AfC CiiC CGT TTC $C GGC TAC ATC QG KVHYLE OONKEIEAEIHALR QKQASHAOLG AAA GTG LX TAC TTG G4A CAA CAG MC MC GAG ATC GAG GCA WIG ATC CAC GCG Clt CGG CAG MC CAG CCC TCG CAC GCC UC Clt ffil D A Y DOEIRELRATLENVNHEKAQVQLDSDH GAC GCT TAC GAC CAG GAG ATC CGA GAG CTG CGC GCC ACC CTG GAG ATG GTG Ml CAC GAG MG GCT CM GTG CAG CTG GAC TCT GA1 CAC EEDIHRLKERFEE R 0 T E A A I R A V R 1:G GAG GM GAC ATC CAC CGG CTC AAG GAG CCC TTC GAG GAG ;G G:G C:G C:G CGG GAC & ACG GAG GCT Ccc ATC CGG GCG GTG XC KDIEESSNVKVELDKKVQS LQDEVAFLRSN AM WC ATA GAG GAG TCG TCG ATG GTT MC GTG GAG CTG GAC MG AAG Gll CAG TCG CTG CAG GAT GAG GTG GCC TTC CTG Cffi AK MT CiC t&l GiG CiG G:G G:C GiC C:G C:G GtC CiG A:C ;G G:G 1:G Cl!& A:C A& G;A it C% Ah CiiC 1:C C:G ;G AL & A,:C Tfc A L K E I R SOLECHSDONHHQAEEN FKCRYA ACG GCG CTG MA GAG ATC CGC TCC CAG CTC GAG TGT C4C TCC GAC CAG MC ATG CAC CAG GCC GAA GAG TGG TTC AAA TGC u;c TAC CCC K L T E A A E Q N K E A I RSAKEEIAEYRRQLQSK AAG CTC ACC GAG GCG GCC C&G CAG MC MG GAG CCC ATC CGC TCC GCT AAA GAA GAG ATC GCC GAG TAC CGG CGC CAG CTG ZAG TCC MG S I E L E S V R G 1 K E S L E S N H D L S AGC All GAG CTC WG TCG GTG CWI GGC ACT MG GAG TCC CTG GAA CiG $G C:C AGC I& A:C CiG l& C:C CiC MC CAC CAC CTC AGC A:C 1:C CiG GiC A:C A:C CiG CiG dg GiA Ail iit C:l C:G Gk AiiA :G T:G ia A?G G:T C:l C!T T:G C& GiA T:C CiG :T C:C LNVKHALDIEIAAYRKLLEGEETRF S T F 5 G Cl1 AAC GTC AAG ATG GCT CTG CAC ATC GAG ATC GCC GCA TAT AGG AAA CTA CTG GAG GGT GAA GAG ACC AG4 TTT AGC AU TTT TCA GGA SITGPLYTHR OPSVTISSKIOKTKVEAPKL AGC ATC ACT GGG CC1 CTG TAC ACA UC CGA CAG CCC TCA GTC ACA ATA TCC AGT AAG All CAG AAG ACC AAA GTC W\G GCC CCC MG CTC QHKFVEEIIE S DALTVI A,, G:C CAA CAC AAA TTT GTG GAG GAG ATC All GAG G:G A:T A:, G:G ;A GiT ;G A:G TCA GiA A:G G:A GAC GCC UC ACA GTC ATT AEELAASAKEEKEEAEEKEEEPEVKSPVKS GCA GAG GAA TTG GCA GCC TCT CCC AAA GAG GAG AAA GAA GAG GCA GAA GAA AAG WV\ GAG GM CCG GAA GTG AAG TCT CCC GTG AAG TCT $1 CiG G:l :G &A $G GkG &A GZ;G &A AiG G:G GiA & CiG GiA G:C ia CiG ;A GiA ;G $G & :l GiA G:l G ;C Ait T: DOAEEGGSEKEGSSEKDEGEOEEEGETEAE GAC CAG GCA GAA GAG GGA GGA TCT GAG AAG GAA GGC TCG AGT GAA AAG GA1 GAA GGT GAG CM GM GAA GAA GGG GM ACT GAG GCA GAA GEGEEAEAKEEK KTEGKVEEMAIKEEIKVE GGT GAA GGA GAG G4A GCA GAA CC1 AAG GAG GAA AAG AAA ACA GAG GGA AAG GTC GAG GM ATG CC1 ATC AAG GAG GM ATC MG GTC GAG P E KSPVPKSPVE E V K P K KAGKDEQ AiG CCC GAG AiA G:C AAG TCC CC1 GTG CCA AAA TCA CCG GTG GAA GAA GTA MG CCA MA CL GiA G& AAA GCC CGA MG CAT GAG CAG E E K K E V E AiG :G CiA GiG ;A Gt;l GAG GAG MG MC GAG GTA G:C AiG GAA l&s C!C ;G Gh :G :G G ;G ;G & AiG GiG :G AG & AiA DVPDKKKAE SP~KEKAVEEMITITK S V K V 5 GA1 GTC CCA GA1 AAA AAG MG GCT GAG TCC CCA GTG AAA GAA MG GCC GTA GAG GM ATG ATC ACC ATT ACT MG TCG GTA MC GTG AGC PQPQEKVKEKA C:G GiG AiA & AiC ia :G :G AiG CC1 CAG CAG CAG GAG MC GTG AAG G4G MG GCA CiG GiG GiG I& Gil A:T CiG :G GiA G:G G D K S-P QESKKEDIAINGEVEGKEE-E Gtl CAC AAA AGC CCG CAA GAA TCC MC MC GM GAC ATA GCT ATC Ml GGG GAG GTG GM GGA MA GAG GAG GAG CiiG CiG 6: Ai:T $G EKGSGQE E E K G V V T N GLDVSPAEEKK G E D II GAG MG GGC AGT GGG CM GAG GAG GAG AM GGG GTG GTC ACT Ml GGC TTA GAT GTG AGC CCT GCG GAG GAA MG AM GGG GAG G4T AG Do V E K I Ail $1 CiC Ah GiG GYG G!G A: AiG GG CTA GM MA ATC A& Ai: GiG ii G!C G.!T GiiT CiT A& Ah T:C A:C A:C AiA T:T G!T TVTQKVEEHEETFE EKLVSTK K VE V T S H A ACT GTC ACT CAA AAG GTT GM GAG CA1 QG GAG ACC TTT GAG GAG MG CTG GTG TCA ACT MA MC GTA GM AiG GTC ACT TCA CAT CCC 840 A:, G:C :G & Gk A:C ZG GET I& TM CATCtCAClCCATTGCMMGtTlMGC~lA~~lTT~TG~TGT~lT~~GCTT~~~l~T TClCCCAlGAGtGCTCCAGA~llGlAlTllCClllGTGCAATATCAGG~CTG~lG~GCTUCGtTCCTCCT~GTCClT~ Figure 3. Complete nucleotide sequence of pnf-m2d and deduced amino acid sequence of rat NF-M. The 5 untranslated region is only 10 base pairs long, whereas the 3 untranslated region is about 150 base pairs long and does not include the poly(a) tail.

5 2594 Napolitano et al. - cdna Cloning of Middle Neurofilament Protein NF-M (Rat) 1 SYTLDSLGNP-SAY-RRVPTETRSSFSRVSGSPSSGFRSQSWSRGSPSTVSSSYKRSALAPRLAYSSAMLSSAESSLDFSQSSSLLNGGSG--GDYKLSRSRE NF-M (Porcine) SYTLDSLGNPSSAY-RRV=TETRSSFSRVSGSPSSGFRSQSWSRGSPSTVSSSYKRSALAPRL~YSSAMLSSAESSLDFSQSSSLLPGGSGPGGDYKLSRSRC NF-L (Mouse) SSFGSDPIFSTSYKRRY-VETPRVHISSVRSGYSTARSAYSS-----YSAPVSSSLSVRRSYSSSSGSLKPSLENLDVSQVAAI SNDLKSIRTQE coil la coil lb NF-M (Rat) 101 KEQLQGLNDRFAGYIEKVHYLEOONKEIEAEIHALRQKQAGHAQLGDAYDQEIRELRATLEMVNHEKAQVQLDSDHLEEDIHRLKERFEEEARLRDDTEA NF-M (Porcine) KEQ~QGLNDRFAGYIEKVHYLEQONKEIEAEI~ALRQKQASHAQLGDAYDQEIRELRATLECVNHEKAQVQLDSDHLEEDIHRLKERFEEEARLRDDTEA NF-L (Mouse) KAQLQDLNHRFASFIERVHELE~NKVLEAELLVLRQKHSEPSRFRALYEQEIRDLRLAAEDATNEKQALQGEREGLEETLRNLQARYEEEVLSREDAEG coil 2 NF-M (Rat) 201 AIRAVRKDIEESSMVKVELDKKVQSLQDEVAFLRSNHEEEVADLLAQIQASHITVERKDYLKTDISTALKEIRSQLECHSDQNMHQLEEWFKCRYAKLTE NF-M (Porcine) AIRACRKDIEEASCVKVELDKKVQSLQDEVAFLRSNHEEEVADLLAQIQASHITVERKDYLKTDIS~ALKEIRSQLECHSDQNM~Q~EEWFKCRYAKLTE NF-L (Mouse) RLMEARKGADEAALARAELEKRIDSLMDEIAFLKKVHEEEIAELQAQIQIAQISVEMDVSSKPDLSAALKDIRAQYEKLAAKNMQNAEEWFKSRFTVLTE NF-M (Rat) 301 AAEQNKEAIRSAKEEIAEYRRQLQSKSIELESVRGTKESLERQLSDIEERHNHDLSSYQDTIQQLENELRGTKWEMARHLREYQDLLNVKMALDIEIAAY NF-M (Porcine) AAEQNKEAIRSAKEEIAEYRRQLQSKSIELESVRGTKESLERQLSDIEERHNHDLSSYQDTIOOLENELRGTKWEMARHLREYQDLLNVKMALDIEIAAY NF-L (Mouse) SAAKNTDAVRAAKDEVSESRRLLKAKTLEIEACRGMNEALEKQLQELEDKQNADISAMQDTINKLENELRSTKSEMARYLKEYHDLLNVKMALDIEIAAY NF-M (Rat) 401 RKLLEGEETRFSTFSGSITGPLYTHRQPSVTISSKIQKTKEAPKLKVQHKFVEEIIEETKVEDEKSEMEDALTVIAEELAASAKEEKEEAEEKEEEPEV NF-M (Porcine) RKLLEGEETRFSTFAGSITGPLYTHRQPSLTISSK FVEEIIEETKVEDEKSEM NF-L (Mouse) RKLLEGEETRLSFTSVGSITSGYSQSSQVFGRSAYSGLQSSYLMSARSFPAYYTSHVQEEQTEVEETIEATKAEEAKDEPPSEGEAEEEEKEKEEGEEE NF-M (Rat) 501 KSPVKSPEAKEEEEGEKEEEEEGQEEEEEEDEGVKSDQAEEGGSEKEGSSEKDEGEQEEEGETEAEGEGEEAEAKEEKKTEGKVEEMAIKEEIKVEKPEK NF-L (Mouse) EGAEEEEAAKDESEDTKEEEEGGEGEEEDTKESEEEEKKEESAGEEQLAKKKD NF-M (Rat) 601 AKSPVPKSPVEEVKPKPEAKGKDEQKEEEKVEEKKEVAKESPKEEKVEKKEEKPKDVPDKK~ESPVKE~VEEMITITKSVKVSLEKDTKEEKP~E NF-M (Rat) 701 KVKEKAEEEGGSEEEVGDKSPQESKKEDIAINGEVEGKEEEEQETQEKGSGQEEEKGVVTNGLDVSPAEEKKGEDRSDDKVVVTKKVEKITSEGGDGATK NF-M (Rat) 801 YITKSVTVTQKVEEHEETFEEKLVSTKKVEKVTSHAIVKEVTQD Figure 4. Comparison of the amino acid sequences of NF-M (rat), NF-M (porcine) from Geisler et al. (1984), and NF-L (mouse) from the deduced amino acid sequence reported by Lewis and Cowan (1986). Residues that differ between the rat and porcine NF-M sequences are underlined in the porcine sequence. Coils la, lb, and 2-which represent the segments of the a-helical rod region of all intermediate filaments-are designated. Mouse NF-L, rather than porcine NF-L (Geisler et al., 1985), is used for the comparison because the mouse sequence is expected to be closer to the rat sequence. distinguishing between NF-M- and NF-H-directed antibodies in unfractionated serum, a result indicating that pnf-m3c codes for NF-M. This evidence, in conjunction with the DNA sequence data described below, forms the basis for our conclusion that pnf-m3c and pnf-m2d are cloned cdna probes encoding the rat NF-M polypeptide subunit. Complete amino acid sequence of rat NF-M The 2 clones were sequenced in both directions, and the location of the 2 cdna clones are shown in Figure 2. The position of the pnf-m3c cdna insert is entirely within the coding region of its mrna and was established by aligning the derived amino acid sequence(s) with the reported sequence of porcine NF-M (Geisler et al., 1984). Direct orientation of the rat and porcine sequences in relation to the structural model of NF-M (Geisler et al., 1983) shows that the cdna encodes a stretch of approximately 500 amino acid residues that begins in the middle of the first a-helical segment (coil la) of the conserved 40,000- molecular-weight rod domain of the molecule and extends into the variable carboxy-terminal tailpiece. Sequence determination of pnf-m2d showed that both the initiation codon and the termination codon are present in the larger of the 2 clones, indicating that pnf-m2d codes for the full-length polypeptide. The nucleotide and the deduced amino acid sequences are shown in Figure 3. As shown in Figure 4, the N-terminal region of the molecule is highly conserved compared with the sequence described by Geisler et al. (1984) for the porcine NF-M. In addition, a high degree of homology between NF-M and NF-L can be discerned. The C-terminal tail contains large stretches of glutamic acid, which was also predicted from the amino acid composition of rat NF-M described by Chiu and Norton (1982). The predicted molecular weight of the molecule is 95,600, considerably smaller than the molecular weight obtained by SDS-PAGE (145 kda for rat NF-M) and about 10% smaller than the molecular weight predicted from the gel-filtration and sedimentation-equilibrium studies of

6 The Journal of Neuroscience, August 1987, 7(8) Sb 18% a b c d e Figure 5. Northern blot analysis of the expression of NF-M mrna transcripts in rat brain during ontogeny. Aliquots oftotal rat brain RNA, 20 rg, were used for hybridization with 32P-labeled pnf-m3c cdna insert. The filter was washed to a final stringency of 0.2 x SSC at room temperature. Lanes a-frepresent PO, P2, P4, P8, P14, and adult, respectively. One brain-specific mrna species of approximately 3000 nt encodes the rat NF-M subunit, which was found to be expressed at a constant level throughout development. The positions of 28s and l&s ribosomal RNA are shown at left. Kaufmann et al. (1984). As described below, pnf-m2d subcloned into pgem could be transcribed and translated into a polypeptide that corn&rated exactly with rat NF-M, indicating that there were no deletions in the cdna that could have resulted in premature termination and, thus, in a truncated protein. Level of expression of NF-A4 mrna during rat brain development Expression of mrna encoding the NF-M subunit during rat brain development was investigated by Northern blotting. Total brain RNA isolated from developmentally staged rats was transferred to appropriate filters and probed with 32P-nick-translated NF-M cdna insert. A single mrna species of approximately 3000 nt (Fig. 5) was detected in brain with pnf-m3c insert labeled with 3ZP by nick-translation (Rigby et al., 1977). The level of NF-M mrna remained constant from postnatal day 0 (PO) up through and including the adult animal (Fig. 5), a finding that corroborates previous studies on the appearance of the NF-M polypeptide subunit during development (Shaw and Weber, 1983; Willard and Simon, 1983; Pachter and Liem, 1984). In vitro transcription/translation NF-M mrna was transcribed in vitro using the pnf-m2d insert subcloned into the pgem system. In order to determine which promoter would direct the synthesis of the sense mrna a b c d Figure 6. In vitro transcription/translation of the pnf-m2d cdna cloned into pgem-2. The transcription was done as described in the text with both the T7 and SP6 promoters. The translations were done in a rabbit reticulocyte lysate system containing?3-methionine. The translation products were separated on 7.5% SDS-PAGE and processed for fluorography. Lane a, negative control (no promoter); lane b, translation products from mrna transcribed with the SP6 promoter; lane c, translation products from mrna transcribed with the T7 promoter; lane d, material from lane c, immunoprecipitated with the anti-neurofilament antibody. The markers denote the position of NF-H (H), NF-M (M), and NF-L (L), which were mn on the same gel. Only a single protein, comigrating with NF-M and immunoprecipitated by the anti-nf-m antibody is synthesized from the in vitro transcribed message using the T7 promoter, indicating that the entire coding region of the protein is present in this cdna clone. strand, separate reactions with SP6 and T7 polymerase were carried out. mrna generated in this fashion was translated in rabbit reticulocytes and the polypeptide products resolved on SDS-PAGE (Fig. 6). The results indicate that NF-M sense mrna was being transcribed from the T7 promoter and upon translation yielded a polypeptide that comigrated with rat NF-M. This translation product was found to be immunoprecipitable with the anti-nf antibody (Fig. 6, lane d). When the immunoprecipitated translated protein was run on 2-dimensional isoelectric focusing (IEF)-SDS-PAGE, the protein showed a similar mobility as NF-M isolated from rat optic nerve (Fig. 7, a, b), indicating that phosphorylation ofrhis protein changes the mobility only slightly, if at all. When the translation was done in the presence of 32P-y-ATP, a large number of the proteins in the rabbit reticulocyte lysate mixture were phosphorylated (Fig. 7~). However, there was no phosphorylated NF-M protein immunoprecipitated by anti-nf (Fig. 7d), indicating that NF-M was not phosphorylated by kinases present in the rabbit reticulocyte lysate system. Discussion In this report, we describe the molecular cloning of a cdna encoding the 145-l 50 kda middle-molecular-weight subunit of rat neurofilaments, NF-M, from a bacteriophage Xgtll expression library. The inducible expression of cloned cdnas in E. coli as fl-galactosidase-nf-m fusion proteins made feasible the application of immunological methods for the detection of NF-encoding clones. A 1.6 kb NF-M cdna isolated in this manner was purified and subcloned into a suitable high-copy plasmid vector for amplification. This cdna was subsequently used to

7 2596 Napolitano et al. * cdna Cloning of Middle Neurofilament Protein Figure 7. Two-dimensional IEF-SDS-PAGE of the in vitro transcription/translation products of the pnf-m2d cdna cloned into pgem described in Figure 6. The ph gradient is from ph 6.5 to 5 (left to right). Lane a, Coomassie blue-stained 2-dimensional gel of the optic nerve cytoskeletal extract, which was coelectrophoresed with the translation product shown in lane b. Lane b,?s-methionine-labeled translation product described in Figure 6, immunoprecipitated with anti-nf. The spot in lane b comigrates with the NF-M spot in the cytoskeletal preparation (arrow, lane a). Lane c, Autoradiogram of the 32P-labeled proteins in the rabbit reticulocyte lysate mixture. Lane d, Autoradiogram of the labeled proteins in lane c immunoprecipitated with anti-nf. No phosphorylated protein corresponding to NF-M was immunoprecipitated. The in vitro translations of the NF-M mrna transcribed from the pnf-m2d clone with 35S-methionine and 3zP were done in parallel. obtain a second, larger cdna (pnf-m2d), which was shown to encode the entire NF-M protein. The complete amino acid sequence of rat NF-M reveals the same structural features already described for the partial amino sequence of porcine NF-M described by Geisler et al. (1984). In particular, the a-helical rod region, which spans residues , is highly conserved and shows 97% homology with the porcine NF-M sequence (Fig. 4). The N-terminal headpiece shows 90% homology with porcine NF-M. The a-helical rod region of rat NF-M also shows 50% homology with mouse NF-L as determined from the nucleotide sequence (Lewis and Cowan, 1985, 1986), as well as porcine NF-L (Geisler et al., 1985). The C-terminal region of NF-M shows a greater variability compared with the C-terminal region of mouse or porcine NF-L. In particular, little homology can be found in the region immediately following coil 2 (residues, of NF-M). The 70 amino acid C-terminal region of mouse NF-L is 40% homologous with rat NF-M (residues ) with a stretch of 12 amino acids that are 75% homologous (residues of NF-M). However, 70% of the amino acids in this C-terminal region of NF-L are glutamic acid residues, and the overall homology may therefore be due partly to the random location of the large number of glutamic acids in these C-terminal extensions. The homologies between rat NF-M and mouse NF-L in their C-terminal extensions do not extend to the C-terminal regions of porcine NF-L and rat NF-M, where the only homologies are between apparently random glutamic acid residues in the 2 sequences. Comparison of the sequence of the C-terminal extension of rat NF-M with NF-M from other species or with NF-H is not possible at this time since no other sequence data are available. However, a subsequence of 39 amino acids reported by Geisler et al. (1984) in the C-terminus of porcine NF-M is not present in rat NF-M. In addition, the predicted structure shows that an additional CNBr fragment would have been observed in rat NF-M as compared with the fragments described by Geisler et al. (1984). The likelihood that the C-terminal

8 The Journal of Neuroscience, August 1987, 7(8) 2597 region of NF-M varies with species is enhanced by the observation that the observed molecular weight of NF-M determined by SDS-PAGE varies with species (Liem et al., 1978; Chiu and Norton, 1982; Shaw et al., 1984). The calculated molecular weight of NF-M is only 65% of the molecular weight predicted by SDS-PAGE and is, in fact, smaller than the molecular weight obtained from the gel filtration and sedimentation equilibrium data of Kaufmann et al. (1984). This result is not completely surprising considering the rather unusual glutamic acid-rich sequences in the C-terminal extension of NF-M. On the basis of the high fraction of glutamic acid (30%), one would predict this C-terminal extension to be a-helical (Chou and Fasman, 1974). However, the long stretches of glutamic acid are likely to result in the formation of an extended helix rather than an a-helix (Maxfield and Scheraga, 1975; Maxfield et al., 1975). Circular dichroism studies of Geisler et al. (1984) have also shown that the C-terminal extension of the NF-M molecule is not a-helical. The presence of a negatively charged extended helix in the C-terminal extension of NF-M could explain the fact that this protein runs anomalously on SDS-PAGE, since it is likely that the extended helix will not turn into a prolate ellipsoid with the standard charge-to-length ratio of other proteins in the presence of SDS. The data of Geisler et al. (1984) have also shown that the C-terminal extension alone is responsible for this anomalously high molecular weight on SDS-PAGE. The stretches of glutamic acid in the C-terminal extension could also be responsible for the difference between the calculated molecular weight and the molecular weight obtained from the gel filtration and sedimentation data of Kaufmann et al. (1984). However, an alternative explanation is that the porcine protein is 10% larger than rat NF-M. Studies on the appearance of NF proteins in rat brain (Shaw and Weber, 1982) and optic nerve (Willard and Simon, 1983; Pachter and Liem, 1984) have revealed a differential expression of the constituent triplet polypeptide subunits during development. NF-L and NF-M, which in rat correspond to the 68,000 and 150,000 NF subunits, respectively, were shown to be present throughout postnatal development, while NF-H, the largest of the 3 NF components emerged much later at postnatal day 20 (P20) in optic nerve (Willard and Simon, 1983; Pachter and Liem, 1984) and was barely discernible before P13 in the developing rat brain (Shaw and Weber, 1982). We therefore examined the levels of NF-M mrna in developmentally staged rats by Northern blotting. mrna encoding NF-M polypeptide was expressed at a constant level throughout the period of postnatal maturation; this result is consistent with a study by Lewis and Cowan (1985), who observed a similar pattern of postnatal expression for the mrna encoding mouse NF-L. Further studies at the genetic level are warranted to determine if there is a coordinate regulation of NF-L and NF-M gene expression. In this context, it is interesting to note that Bennett and DiLullo (1985) have reported the possible appearance of NF-M prior to vimentin and NF-L in some chick neuroepithelial cells, which would argue against coordinate expression. The calculated isoelectric point of rat NF-M is 5.55, which is not far from the observed isoelectric point on 2-dimensional gels (Pachter and Liem, 1984). The calculated p1 obviously does not take into consideration the charge-charge interactions, which may occur in the carboxy-terminal extension of the molecule, which is very rich in glutamic acid and lysine. When the in vitro translated material was run on l- and 2-dimensional gels, the mobility of the protein was similar to that observed for native NF-M (Figs. 6 and 7). The change in p1 of this protein due to phosphorylation is apparently relatively small, in agreement with the results of Black et al. (1986), who showed that posttranslational modification of NF-M in cultured cells caused a small shift in the p1. These small shifts could easily have been missed in our experiments, since we found that the in vitro translated protein ran as a relatively elongated spot. The number of phosphate groups on NF-M have been reported to be 9-25 on the basis of a molecular weight of 150,000 (Jones and Williams, 1982; Julien and Mushynski, 1982; Wong et al., 1984; Carden et al., 1985). Since the molecular weight is, in fact, 95,600, the actual number would be 6-l 5. This number of phosphate groups on the NF-M molecule would lower the p1 of the molecule by 0.05-o. 10 ph units after phosphorylation. The lack of phosphate groups on the in vitro translated protein was shown also by doing the translation in the presence of 32P-ATP, followed by immunoprecipitation of the translation product. No phosphorylated material was immunoprecipitated, indicating that kinases in the reticulocyte lysate preparation were not responsible for the phosphorylation of the protein. These studies also argue against the possibility that the S 150 protein observed in our laboratory (Wong et al., 1984) is solely an unphosphorylated form of NF-M, since this protein is considerably more basic than the unphosphorylated protein. Other posttranslational modifications of this protein are therefore possible, since the S 150 protein was recognized by highly specific antibodies to NF-M. Therefore, the fact that the S150 protein did not copolymerize with NF-L cannot be due solely to the low number of phosphate groups on this protein, in agreement with the results of Georges et al. (1986) who showed that enzymatically dephosphorylated NF-M and NF-H could still copolymerize with NF-L into a filament. Although little is known about the possible phosphorylation sites present in the NF-M molecule, it is interesting to note that the sequence lysine-serine-proline (KSP) appears 5 times in the C-terminal extension of NF-M. There are 3 additional serineproline (SP) sequences in the C-terminal extension and 2 in the N-terminal head region of NF-M. If these sequences represent possible phosphorylation sites, they would be sufficient to account for most of the phosphate groups on rat NF-M. In addition, none of these sequences is also present in NF-L, which is known to be phosphorylated to a much lesser extent than NF-M. When more phosphorylated proteins in the nervous system, as well as in other cell types, are analyzed, it will be of interest to determine if similar sequences are present. In particular, NF-H is known to be even more heavily phosphorylated than NF-M, and if these sequences are in fact phosphorylation sites, we would expect them to occur with much greater frequency. While the establishment of NF-H as an NF accessory protein has been substantiated by immunoelectron microscopy (Willard and Simon, 1981; Debus et al., 1982; Sharp et al., 1982; Hirokawa et a1.,1984; Liem et al., 1985) the precise disposition of NF-M in relation to the core polymer remains somewhat equivocal. The long carboxy-terminal domains of NF-H and NF-M most likely confer upon them their immunologic as well as functional specificity in nervous tissue. Assembly studies have shown the ability of NF-L to reassemble into intermediate-sized filaments (Geisler and Weber, 198 1; Liem and Hutchison, 1982), whereas NF-M and NF-H are able to polymerize into filaments only under highly specified experimental conditions. The unusual C-terminal sequence of NF-M could provide an explanation for this difficulty in assembly, since electrostatic repul-

9 2599 Napolitano et al. * cdna Cloning of Middle Neurofilament Protein sion would occur between adjacent C-terminal tail vieces. These intermediate filaments into epithelial and non-epithelial members. electrostatic interactions could be diminished by the addition EMBO J. 2: Geisler, N., E. Kaufmann, S. Fischer, U. Plessmann, and K. Weber of salt, which has been shown to favor the assembly of neuro- (1983) Neurofilament architecture combines structural principles of filaments (Liem and Hutchison, 1982). intermediate filaments with carboxy-terminal extensions increasing With the acquisition of full-length NF-encoding cdna clones, in size between triolet nroteins. EMBO J. 2: : we will be afforded the means to develop specific probes in appropriate eukaryotic vectors for the investigation of NF function and expression in vitro and in vivo. Such biological tools may also be utilized to study the assembly and phosphorylation ofthe NF proteins. Genomic clones can be used to study neuronspecific promotors, and the expression of these proteins in nonneuronal cells will allow us to study the possible function of each of the NF triplet proteins. NF-M, which is one of the major neuron-specific proteins, has not yet been assigned a function, and the availability of cdnas for this protein should aid us in determining this function. References Aviv, H., and P. Leder (1972) Purification of biologically active globin messenger RNA by chromatography on oligothymidylic acid cellulose. Proc. Natl. Acad. Sci. USA 69: Bennett, G. S., and C. DiLullo (1985) Transient expression of a neurofilament protein by replicating neuroepithelial cells of the embryonic chick brain. Dev. Biol. 107: Biggin, M. D., T. J. Gibson, and G. F. Hong (1983) Buffer gradient gels and [%I label as an aid to rapid DNA sequence determination. Proc. Natl. Acad. Sci. USA 80: Black, M. M., P. Keyser, and E. Sobel (1986) Interval between the synthesis and assembly of cytoskeletal proteins in cultured neurons. J. Neurosci. 6: Bloemendal, H., W. Quax, Y. Quax-Jeuken, R. Van den Heuvel, W. V. Egberts, and L. Van den Broek (1985) Vimentin and desmin cdna clones: Structural aspects of corresponding proteins and genes. Ann. NY Acad. Sci. 455: Boedtker, H. (197 1) Conformation independent molecular weight determination of RNA by gel electrophoresis. Biochim. Biophys. Acta 240: Carden, M. J., W. W. Schlaepfer, and V. M-Y. Lee (1985) The structure, biochemical properties and immunogenicity of neurofilament peripheral regions as determined 260: by phosphorylation. J. Biol. Chem. Chirgwin, J. M., A. E. Przybyla, R. J. MacDonald, and W. J. Rutter (1979) Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease.biochemistry 18: Chiu. F. C.. and W. T. Norton (1982) Bulk oreoaration of CNS cvtoskeleton and the separation of individual^neurofilament proteins by gel filtration: Dye binding and amino acid composition. J. Neurochem. 39: Chou, P., and G. D. Fasman (1974) Conformational parameters for amino acids in helical, beta-sheet, and random coil calculated from proteins. Biochemistry 12: 2 1 l-222. Dale, R. M. K., B. A. McClure, and J. P. Houchins (1985) A rapid single-stranded cloning strategy for producing a sequential series of overlapping clones for use in-dna- sequencing: Application to sequencing the corn mitochondrial 18s rdna. Plasmid 13: 3 l-40. Debus, E., G. Flugge, K. Weber, and M. Osbom (1982) A monoclonal antibody specific for the 200K polypeptide of the neurofilament triplet. EMBO J. 1: Gardner, E. E., D. Dahl, and A. Bignami (1984) Formation of 10 nm filaments from the 150k-dalton rosci. Res. 11: neurofilament protein in vitro. J. Neu- Geisler, N., and K. Weber (198 1) In vitro self-assembly of the 68,000 molecular weight component ofthe mammalian neurofilament triplet proteins into intermediate-sized 571. filaments. J. Mol. Biol. 151: 565- Geisler, N., and K. Weber (1982) The amino acid sequence of chicken muscle desmin provides a common structural model for intermediate filament oroteins. EMBO Geisler, N.,* and K. Weber. J. 1: (1983) Amino acid sequence data on glial fibrillary acidic protein (GFA); Implications for the subdivision of Geisler, N., S. Fischer, J. Vandekerckhove, U. Plessmann, and K. Weber (1984) Hybrid character of a large neurofilament protein (NF-M): Intermediate filament type sequence followed by a long and acidic carboxy-terminal extension. EMBO J. 3: Geisler, N., U. Plessmann, and K. Weber (1985) The complete amino acid sequence of the major mammalian neurofilament protein (NF- L). FEBS Lett. 182: Georges, E., S. Lefebvre, and W. E. Mushynski (1986) Dephosphorylation of neurofilaments by exogenous phosphatases has no effect on reassembly of subunits. J. Neurochem. 47: Goldman, B. M., and G. Blobel (1978) Biogenesis of peroxisomes: Intracellular site of synthesis of catalase and &case. Proc. Natl. Acad. Sci. USA. 75: Gubler, II., and B. J. Hoffman (1983) A simple and very efficient method for generating cdna libraries. Gene 25: : Hanukoglu, I., and E. Fuchs (1982) The cdna sequence of a human epidermal keratin: Divergence of sequence but conservation of structure among intermediate filament proteins. Cell 31: Hanukoglu, I., and E. V. Fuchs (1983) The cdna sequence of a type II cytoskeletal keratin reveals constant and variable structure domains among keratins. Cell 33: Hirokawa, N., M. A. Glicksman, and M. B. Willard (1984) Organization ofmammalian neurofilament polypeptides within the neuronal cytoskeleton. J. Cell Biol. 98: Hoffman, P. N., and R. J. Lasek (1975) The slow component of axonal transport. Identification of major structural polypeptides of the axon and their generality among mammalian neurons. J. Cell Biol. 66: Hoffman, P. N., J. W. Griffin, and D. L. Price (1984) Control ofaxonal caliber by neurolilament transport. J. Cell Biol. 99: Jones, S. M., and R. C. Williams (1982) Phosphate content of mammalian neurofilaments. J. Biol. Chem. 257: Julien, J. P., and W. E. Mushynski (1982) Multiple phosphorylation sites in mammalian neurofilament polypeptides. J. Biol. Chem. 257: Kaufmann, E., N. Geisler, and K. Weber (1984) SDS-PAGE strongly overestimates the molecular masses of the neurofilament proteins. FEBS Lett. 170: 8 l-84. Laskey, R. A., and A. D. Mills (1975) Quantitative film detection of 3H and 14C in polyacrylamide gels by fluorography. Eur. J. Biochem. 56: Lazarides, E. (1982) Intermediate filaments. Annu. Rev. Biochem. 51: Lewis, S. A., and N. J. Cowan (1985) Genetics, evolution, and expression ofthe 68,000-mol-wt neurofilament protein: Isolation ofa cloned cdna probe. J. Cell Biol. 100: Lewis, S. A., and N. J. Cowan (1986) Anomalous placement of introns in a member of the intermediate filament multigene family: An evolutionary conundrum. Mol. Cell. Biol. 6: 1529-l 534. Lewis, S. A., J. M. Balcarek, V. Krek, M. L. Shelanski, and N. J. Cowan (1984) Sequence of a cdna clone encoding mouse glial fibrillary acidic protein: Structural conservation of intermediate filaments. Proc. Natl. Acad. Sci. USA 81: Liem. R. K. H.. and S. B. Hutchison (1982) Purification of individual components of the neurofilament triplet: Filament assembly from the 70,000-dalton subunit. Biochemistry 21: Liem, R. K. H., S. H. Yen, G. D. Salomon, and M. L. Shelanski (1978) Intermediate filaments in nervous tissues. J. Cell Biol. 79: Liem, R. K. H., S. S. M. Chin, E. Moraru, and E. Wang (1985) Monoclonal antibodies to epitopes on different regions of the 20,000 dalton neurofilament protein. Exp. Cell Res. 156: Maxfield, F. R., and H. A. Scheraga (1975) The effects of neighboring charges on the helix forming ability of charged amino acids in proteins. Macromolecules 8: Maxfield, F. R., J. E. Alter, G. T. Taylor, and H. A. Scheraga (1975) Helix-coil stability constants for the naturally occurring amino acids in water. IX. Glutamic acid parameters from random poly(hydroxybutylglutamine)-co-l-glutamic acid. Macromolecules 8:

10 The Journal of Neuroscience, August 1987, 7(8) 2599 Melton, D. A., P. A. Krieg, Rebaghati, T. Maniatis, K. Zinn, and M. edness of neurofilament proteins of higher vertebrates. Eur. J. Cell R. Green (1984) Efficient in vitro synthesis of biologically active Biol. 34: RNA and RNA hybridization probes from plasmids containing a Steinert, P. M., R. H. Rice, D. R. Roop, B. L. Trus, and A. C. Steven bacteriophage SP6 promoter. Nucleic Acids Res. 12: (1983) Complete amino acid sequence of a mouse epidermal keratin Messing, J., and J. Vieira (1982) A new pair of Ml3 vectors for subunit and implications for the structure of intermediate filaments. selecting either DNA strand of double-digest restriction fragments. Nature 302: Gene 19: Tokutake, S., R. K. H. Liem, and M. L. Shelanski (1984) Each com- Morris, J. R., and R. J. Lasek (1982) Stable polymers of the axonal cytoskeleton: The axoplasmic ghost. J. Cell Biol. 92: 192-l 98. Ngai, J., Y. G. Capetanaki, and E. Lazarides (1985) Expression of the genes coding for the intermediate filament proteins vimentin and desmin. Ann. NY Acad. Sci. 455: Pachter, J. S., and R. K. H. Liem (1984) The differential appearance of neurofilament triplet polypeptides in the developing rat optic nerve. Dev. Biol. 103: Quax-Jeuken, Y., W. J. Quax, and H. Bloemendal (1983) Primary and secondary structure of hamster vimentin predicted from the nucleotide sequence. Proc. Natl. Acad. Sci. USA 80: Riebv. P. W. J.. M. Dieckman. C. Rhodes. and P. Bera f 1977) Labeling deoxyribonucleic acid to high specific activity in k& by nick translation with DNA polymerase I. J. Mol. Biol. 113: Sanger, F., A. R. Coulsen, B. G. Barrel, J. H. Smith, and B. Roe (1980) Cloning in single-stranded bacteriophage as an aid to rapid DNA sequencing. J. Mol. Biol. 143: Sharp, G., G. Shaw, and K. Weber (1982) Immunoelectron microscopical localization of the three neurofilament triplet proteins along neurofilaments of cultured dorsal root ganglion neurones. Exp. Cell Res. 137: Shaw, G., and K. Weber (1982) Differential expression of neurofilament triplet proteins in brain development. Nature 298: Shaw, G., E. Debus, and K. Weber (1984) The immunological relat- ponent of neurofilament assembles itself to make component-specific filament. Biomed. Res. 5: Towbin, H., T. Straelin, and J. Gordon (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose: Procedure and some annlication. Proc. Natl. Acad. Sci. USA 76: Weinberg&, C., S. M. Hollenberg, E. S. Ong, J. M. Harmon, S. T. Browser, J. Cidlowski, E. B. Thompson, M. G. Rosenfeld, and R. M. Evans (1985) Identification of human glucocorticoid receptor complementary DNA clones by epitope selection. Science 228: Willard, M. B., and C. Simon (198 1) Antibody decoration of neurofilaments. J. Cell Biol. 89: Willard, M. B., and C. Simon (1983) Modulations of neurofilament axonal transport during the development of rabbit retinal ganglion cells. Cell 35: 55 l-559. Wong, J., S. B. Hutchison, and R. K. H. Liem (1984) An isoelectric variant of the 150,000-dalton neurofilament polypeptide: Evidence that phosphorylation state affects its association with the filament. J. Biol. Chem. 259: Young, R. A., and R. W. Davis (1983) Efficient isolation of genes by using antibody probes. Proc. Natl. Acad. Sci. USA 80: 1194-l 198. Zehner, Z. E., and B. M. Paterson (1983) Characterization of the chicken vimentin gene: Single copy gene producing multiple mrnas. Proc. Natl. Acad. Sci. USA 80: